¨ A
. Demirgu c¸-Kunt, H. Huizinga Journal of Public Economics 79 2001 429 –453 431
abundant opportunities to shift profits either to their home country or to a tax haven, with the expected result that taxes paid relative to the scale of activities as
measured by assets are lower. The raw data indeed suggest that taxes paid by foreign banks are relatively low in many of the major industrialized nations.
Differences in taxes paid, however, may also reflect different operations and activities rather than only profit shifting. Therefore, it is useful to consider taxes
paid, while controlling for indicators of the activity mix as derived from bank- level accounting data. For many developed countries, the evidence again suggests
that foreign banks pay relatively low taxes. Further support for the profit shifting hypothesis is obtained by examining the relationship between taxes paid and the
statutory tax rate. A negative relationship between taxes paid and the statutory tax rate suggests the presence of profit shifting. Our estimates suggest that the
relationship between taxes paid and the statutory tax rate is positive for domestic banks, but negative for foreign banks. This we interpret as evidence of profit
shifting by foreign banks.
This paper contributes to a small empirical literature on profit shifting, all related to non-financial U.S. multinationals. Grubert and Mutti 1991 demonstrate
that foreign affiliates of U.S. multinationals are more profitable in host countries with relatively low levels of taxation, as evidence of the presence of profit shifting.
Along similar lines, Hines and Rice 1994 find that measures of U.S. multination- als’ pre-tax profitability are negatively related to host country average tax rates.
Harris et al. 1993 focus on the U.S. tax liability of U.S. multinational manufacturing firms with foreign subsidiaries. These authors find that U.S. firms
with subsidiaries in low-tax countries report relatively low U.S. tax liabilities, again supporting the profit shifting hypothesis. Finally, Grubert and Slemrod
1998 consider the joint decision of a U.S. multinational to locate in Puerto Rico and whether to shift profits to Puerto Rico. Puerto Rico is found to be especially
attractive to firms with considerable intangible assets that can benefit from income shifting opportunities.
The remainder of this paper is organized as follows. Section 2 presents a concise theoretical model of the relationships between reported profitability, taxes paid and
the statutory tax rate for domestic and foreign banks. Foreign banks may receive foreign tax credits for local taxed paid, and may or may not engage in outward
profit shifting. Section 3 discusses the data. Section 4 presents the empirical results, while Section 5 concludes.
2. The model
This section presents a theoretical model outlining how reported profits and taxes paid are related to the de jure tax rate on banks. Three types of banks are
distinguished. First, there are purely domestic banks. Second, there are foreign banks that are able to obtain a possibly partial foreign tax credit for local taxes
¨ 432
A . Demirgu c¸-Kunt, H. Huizinga Journal of Public Economics 79 2001 429 –453
from their home-country tax authority, and that do not engage in any profit shifting. Third, there are foreign banks that receive a foreign tax credit and in
addition have profit shifting opportunities. The three types of banks are considered in turn.
2.1. A domestic bank Consider a domestic firm that makes pre-tax profits divided by assets equal to p.
The domestic tax rate applied to these profits is denoted t. After-tax profits p1 2 t have to be sufficient to satisfy an international equity holder in the bank
who can obtain an alternative nominal return on his investment i . Therefore
after-tax profitability has to satisfy the following arbitrage relationship p 1 2 t 5 ie
1
1
where e is equity divided by assets. Eq. 1 straightforwardly implies that p is positively related to e
, i and t, as dp dt 5 p 1 2 t . 0. Actual taxes paid, or tp, are also positively related to t, as dtp dt 5 p 1 2 t . 0. The required nominal
return i should be taken to reflect inflation as well as investor risk as proxied by several bank and macro variables in the later empirical work.
2.2. A foreign bank that receives a foreign tax credit The foreign bank, as before, pays tax at a rate t in the domestic or source
country. In its country of residence, the foreign bank additionally pays tax at a rate t, but there the foreign bank also receives a foreign tax credit equal to a share a
of source-country taxes with 1 a 0. The arbitrage relationship for after-tax profitability in 1 is now amended as follows
ˆ p1 2 t 5 ie
2 ˆ
with t 5 1 2 at 1 t. In Eq. 2, pre-tax profits p remain positively related to e and i
, while they are positively related to the domestic tax rate t if a , 1, as then ˆ
dp dt 5 1 2 ap 1 2 t . 0. Actual domestic taxes remain positively related to the domestic tax rate t, as
1 2 t ]]
dtp dt 5 p . 0
F G
ˆ 1 2 t
Note that pre-tax profits and domestic taxes paid are less sensitive to t for a foreign bank than for a domestic bank if a . t, and vice versa. With significant
1
The model abstracts from equity capital gains.
¨ A
. Demirgu c¸-Kunt, H. Huizinga Journal of Public Economics 79 2001 429 –453 433
foreign tax credits available, pre-tax profits and taxes paid are thus relatively insensitive to the domestic tax rate for foreign banks.
2.3. A foreign bank with profit-shifting opportunities Finally, we consider a foreign bank that faces a tax environment as before, with
the only difference that now the foreign bank can shift part of its profits to the
2
parent company. Clearly, the purpose of such profit shifting is to reduce the banking firm’s worldwide tax liability. In practice, there are several avenues for an
international banking firm to shift profits among its national constituents. For instance, a national subsidiary can shift profits outward by paying a relatively high
interest rate on an intracompany loan received from the parent company. Alternatively, the subsidiary can be overcharged for any other non-financial
service provided by the parent company to the subsidiary. At the same time, the international banking firm can reduce a particular subsidiary’s reported profits by
burdening it with a relatively high share of badly performing loans.
With profit shifting, a distinction must be made between ‘true’ profits, p , and
r
reported profits p . Note that reported profits in the source country attract an ˆ
effective tax t, while profits shifted to the parent company are only taxed at a rate ˆ
t , t. Thus profit shifting saves tax but, we assume, it also involves some costs. These costs, denoted c, are taken to be convex in the amount of profits shifted
r
equal to p 2 p , with c0 5 0, c9 . 0, c0 . 0. The arbitrage relation for after-tax profitability now becomes
r r
p1 2 t 2 p 1 2 at 2 cp 2 p 5 ie 3a
r
The firm chooses reported profits, p , so as to maximize worldwide after-tax and after-cost profits on the left-hand side of 3a, given actual profits p
. This yields the following optimality condition
r
c9p 2 p 5 1 2 at 3b
equating the marginal cost of profit shifting to its tax benefit. From 3a and 3b, we can derive how a change in the domestic tax rate t
r
affects actual profits, p, reported profits, p , and taxes paid in the domestic
r
country, tp . Actual profits, p, are easily shown to increase with t for a , 1, as
2
Alternatively, we can assume that the foreign bank can shift profits outward to a tax haven with zero taxes. Eqs. 3a and 3b then become
r r
ˆ p 2 p t 2 cp 2 p 5 ie
r
ˆ c9p 2 p 5 t
r r
yielding exactly the same expressions for dp dt and dtp dt.
¨ 434
A . Demirgu c¸-Kunt, H. Huizinga Journal of Public Economics 79 2001 429 –453
r r
ˆ then dp dt 5 p 1 2 a 1 2 t . 0. Next, reported profits, p , are ambiguously
related to the domestic tax rate t, as
r
p 1
r
]] ]
dp dt 5 1 2 a 2
_ 0.
F G
ˆ c0
1 2 t Similarly, actual and reported taxes in the source country vary ambiguously with
t , as
1 2 t t1 2 a
r r
]] ]]]
dtp dt 5 p 2
_ 0.
F G
ˆ c0
1 2 t These two ambiguous relationships reflect that a domestic tax rate increase entails
the need to generate higher pre-tax profits to pay these taxes, but it also increases
r
the incentive to shift profits outward. It can further be checked that p and p both increase with i and e. Note that for all three banking cases considered, actual
domestic profits and the domestic tax rate t can be positively related. A negative relationship between these two variables, however, is only possible for the third
case of a foreign bank with profit shifting opportunities.
3. The data
